Envlron. Sci. Technol. 1992, 26, 2395-2408
(16) Thornton, T. D.; Savage, P. E. Proc. Int. Conf. Supercrit. Fluids, 2nd 1991, 421. (17) Thornton, T. D.; Savage, P. E. AIChE J. 1992,38, 321. (18) Li, R.; Savage, P. E.; Szmukler, D. AIChE J., in press. (19) Dean, J. A., Ed. Lunge's Handbook of Chemistry, 12th ed.; McGraw-Hik New York, 1979. (20) Thornton, T. D. Ph.D. Diseertation, University of Michigan, 1991. (21) Thornton, T. D.; Savage, P. E. J . Supercrit. Fluids 1990, 3, 240. (22) Thornton, T. D.; LaDue, D. E., 111; Savage, P. E. Environ. Sci. Technol. 1991, 25, 1507. (23) Fogler, H. S. Elements of Chemical Reaction Engineering, 2nd ed.; Prentice-Hall: Englewood Cliffs, NJ, 1992. (24) Thornton, T. D.; Savage, P. E. Ind. Eng. Chem. Res., in press. (25) Li, L.; Chen, P.; Gloyna, E. F. AIChE. J 1991,37, 1687.
Thomason, T. B.; Modell, M. Hazard. Waste 1984,1,453. Modell, M. U.S. Patent No. 4,338,199, July 6, 1982. Modell, M. U.S.Patent No. 4,543,190, Sept 24, 1985. Helling, R. K.; Tester, J. W. Energy Fuels 1987, 1, 417. Helling, R. K.; Tester, J. W. Environ. Sci. Technol. 1988, 22, 1319. Webley, P. A.; Tegter, J. W.; Holgate, H. R. Ind. Eng. Chem. Res. 1991, 30, 1745. Webley, P. A.; Tester, J. W. In Supercritical Fluid Science and Technology;Johnston, K. P., Penninger,J. M. L., Eds.; ACS Symposium Series 406; American Chemical Society: Washington, DC 1989; pp 259-275. Webley, P. A.; Tester, J. W. Energy Fuels 1991, 5 , 411. Holgate, H. R.; Tester, J. W. Proc. Int. Conf. Supercrit. Fluids, 2nd 1991, 177. Rofer, C. K.; Streit, G. E. Los Alamos National Laboratory Report, LA-11439-MS, DOE/HWP-64 1988. Rofer, C. K.; Streit, G. E. Los Alamos National Laboratory Report, LA-11700-MS, DOE/HWP-90, 1989. Wightman, T. J. M.S. Thesis, University of California at Berkeley, 1981. Yang, H. H.; Eckert, C. A. Ind. Eng. Chem. Res. 1988,27, 2009.
Received for review March 23,1992. Revised manuscript received July 30, 1992. Accepted August 3, 1992. This project was supported by the National Science Foundation (CTS-8906860 and CTS-9015738).
Respeciation of Organic Gas Emissions and the Detection of Excess Unburned Gasoline in the Atmosphere Robert A. Harley, Mlchael P. Hannlgan, and Glen R. Cam"
Environmental Engineering Science Department, Callfornia Institute of Technology, Pasadena, Callfornia 91 125 The development of a set of organic gas composition profiles for key source categories is described. This information is used to recompute the organic gas emission inventory for the Los Angelea area. Comparisons are made between the revised emission inventory and ambient concentration measurements in southern California. Respeciation of the organic gas emissions results in large changes in the basinwide emissions estimates for many individual organic species, including l,&butadiene, ethylene glycol, methanol, and cyclohexane. Significant changes are observed in the reactivity of the chemical composition profiles for individual source categories, especially for surface-coating activities and associated thinning solvent use. Receptor-modeling methods are used to identify the relative importance of major sources that contribute to atmospheric organic gas concentrations in southern California. The receptor modeling results indicate a key discrepancy between the emisaion inventory and ambient data: there is much more unburned gasoline in the atmosphere than is indicated in the emission inventory. These excess unburned gasoline emissions may be coming from a combination of sources including tailpipe emissions, hot-soak evaporative emissions, and fuel spillage. ~~
1 . Introduction
Preparation of accurate speciated organic gas emission inventories is necessary for photochemical modeling calculations and for the design of ozone abatement strategies. Knowledge of organic gas emissions is also required if the concentrations of toxic air contaminants (e.g., formaldehyde, benzene, and 1,3-butadiene) are to be controlled in a systematic fashion. Current emission inventories in use in the Los Angeles area specify temporally and spatially resolved organic gas emissions for over 800 source Categories. These inventories have been used in the formulation of pollutant abatement 0013-936X/92/0926-2395$03.00/0
strategies (I). Because the Los Angeles ozone control problem plays a critical role in establishing California and nationwide emission control policies, a correct understanding of the Los Angeles area organic gas emission inventory is very important. An organic gas emission inventory combines source activity factors with pollutant emission factors and speciation profiles. For example, detailed vehicle exhaust emissions are calculated as the product of an activity factor (vehicle miles traveled), emission factors (total organic gas, oxides of nitrogen, and carbon monoxide mass emission rates per mile traveled), and speciation factors for organic gases and oxides of nitrogen (percent by weight of individual compounds). Many of the same chemical composition profiles for organic gas emissions used in the California emission inventory have been incorporated into the United States Environmental Protection Agency (EPA) volatile organic compound speciation data system (2). In light of recent studies of motor vehicle emissions on the road, a great deal of attention has been focused on mobile source emission factors (3-5). Measurements made in a roadway tunnel in the Los Angeles area suggest that organic gas emissions from vehicles in actual use on the road are up to 3 times higher than those predicted by the EMFAC 7E mobile source emissions model (6). The EMFAC model is a central part of all mobile source emission calculations for regulatory planning purposes in California A comparison of emission inventory and ambient concentration ratios of carbon monoxide, oxides of nitrogen, and non-methane organic gases (NMOG) also suggests that on-road vehicle emissions of CO and NMOG are understated by the EMFAC model (7).Resolving the emission factor questions is an important step, but both source activity and speciation profile components of the emission calculation also must be considered. In the present study, existing speciation profiles for organic gas emissions will be reviewed, and new or updated
0 1992 American Chemical Society
Environ. Scl. Technol., Vol. 26, No. 12, 1992 2395
Table I. Non-Methane Organic Gas Emissions Assigned to Selected Speciation Profiles in the Emission Inventory for August 27, 1987
profile description" noncatalyst gasoline engine exhaust catalyst-equipped gasoline engine exhaust composite industrial surface coatings solvent-borne architectural surface coatings gasoline vapors whole gasoline domestic solvents industrial adhesives species unknown diesel engine exhaust composite thinning solvent water-borne architectural surface coatings aerosol propellants all other anthropogenic total anthropogenic
non-methane organic gas emissions (tons/day) official revised
346 256 161 120 57 104 49 57 94 44 0 25 57 419 1789
266 246 206 121 115 108 90 57 49 39 30 25 16 412 1780
Note that each speciation profile may be used to speciate organic gas emissions from many different sources. For example, the noncatalyst gasoline engine exhaust profile is used to speciate emissions from on-road vehicles, off-road vehicles, heavy-duty eauipment, and various other sources.
information will be used to improve the accuracy and level of chemical detail of these profiles. The main motivation for this study is to improve the emission estimates for individual organic species, to provide the basis for detailed modeling of the emissions to ambient air quality relationship for the many individual organic species. It is also desirable to check the official state of California emission inventory to determine whether the chemical speciation (and therefore reactivity) of organic gas emissions is defined correctly. For this purpose, a reactivity scale based on reaction rate constants of organic gases with the hydroxyl radical will be used to compare the photochemical reactivity of the revised speciation profiles with the corresponding speciation profiles in the official emission inventory. Our revised estimate of the composition of basinwide organic gas emissions then will be compared with ambient organic gas concentration measurements made in the Los Angeles area during recent field studies (8, 9).
2. Speciation Profile Development The South Coast Air Quality Management District and the California Air Resources Board have developed official emission inventories for the South Coast air basin. A copy of the current emission inventory for August 27,1987, was received from the California Air Resources Board (IO),and forms the starting point for the present study. In Table I, the non-methane organic gas emissions associated with selected speciation profiles are presented. Note that emissions from multiple source categories may be assigned to a single speciation profile. Inspection of Table I indicates that 77% of the anthropogenic nonmethane organic gas emissions is assigned to only 13 speciation profiles. Most of the speciation profiles listed in Table 1 will be revised in the course of this study, as discussed in the following sections. The most important of these revised speciation profiles are shown in Tables I1 and 111. A hydroxyl radical reactivity factor ( K j )has been computed for each speciation profile j , stated relative to a unit 2308
Environ. Sci. Technol., Voi. 26, No. 12, 1992
mass of organic gas emissions:
Kj = C(wi,/M;)koHi i=l
(1)
where wij is the weight fraction of species i in speciation profile j , Mi is the molecular weight of species i , and kOHi is the second-order hydroxyl radical reaction rate constant for species i. In this paper, all reactivities will be expressed relative to the reactivity of the composite organic gas emissions from all sources combined in the official basinwide emission inventory:
CE; i=l
where E; is the basinwide emission rate for species i in the official inventory. The reactivity index is computed for each speciation profile as The reactivity scale defiied above accounts for differences in the reactivity of individual organic species with the hydroxyl radical, and also for variations in molecular weights leading to different molar emissions at the same total mass emission rate. The values for kOHreported by Carter (11)were used in these calculations. There are other reactivity scales that measure the effect on ozone formation of an incremental addition of a single organic species to an existing mixture of oxides of nitrogen and organic gases (12). These incremental reactivity scales consider mechanistic as well as kinetic properties of the species in the assessment of reactivity. As well, the incremental reactivity scales account for photolysis of species such as formaldehyde and reactions with other radicals and ozone. However, the detailed chemical reaction mechanisms of many species are still uncertain, while kOHvalues are well-known. Therefore, the Rj index will be used to illustrate differences in reactivity among various speciation profiles, without assuming a particular ambient hydrocarbon-NO, ratio. 2.1. Mobile Source Evaporative Emissions. The composition of whole (liquid) gasoline and gasoline headspace vapors was investigated by Oliver and Peoples (13). Composites of fuel samples obtained from all of the major oil companies selling gasoline in the Los Angeles area in 1984 were analyzed. Separate analyses were performed for leaded and unleaded fuels, for regular and premium grades, and for summertime and wintertime fuels. The existing Los Angeles emission inventory uses a composite speciation profile that was back-calculated to reflect the sales of gasoline by grade as they existed in 1979. Since unleaded gasoline sales have increased greatly with the further introduction of catalyst-equipped cars into the vehicle fleet, it is necessary to update the weightings applied to the composition profiles for the different grades of gasoline to reflect 1987 sales (14). Use of the 1987 grade splits shown in Table IV results in a composite whole gasoline speciation profie with higher aromatic and lower alkane content. The effect is to increase the reactivity of whole gasoline emissions by 4%. Gasoline vapors that exist in the headspace over the liquid in a gasoline tank are dominated by the higher vapor pressure compounds in the fuel (i.e., butane and pentane). When this gasoline vapor composition is updated to reflect 1987 fuel sales, the reactivity of the gasoline vapors is seen to increase by 9% relative to the vapor composition for 1979 that is still used
Table 11. Revised Speciation Profiles for Exhaust and Fuel Evaporative Emissions
chemical species methane ethane propane n-butane isobutane n-pentane isopentane n-hexane 2-methylpentane 3-methylpentane 2,2-dimethylbutane 2,3-dimethylbutane n-heptane branched C7 alkanes n-octane branched C8 alkanes C9 alkanes C10 alkanes C11 alkanes C12+ alkanes cyclopentane methylcyclopentane cyclohexane methylcyclohexane other cycloalkanes acetylene ethene propene 1-butene isobutene cis-2-butene trans-2-butene C5 terminal alkenes C5 internal alkenes C6 terminal alkenes C6 internal alkenes C7 alkenes C8 alkenes C9+ alkenes cyclopentene other cycloalkenes 1,3-butadiene other dialkenes benzene toluene ethylbenzene other monoalkylbenzenes o-xylene m- and p-xylene ethyltoluene isomers diethylbenzene isomers 1,3,5-trimethylbenzene 1,2,44rimethylbenzene 1,2,3-trimethylbenzene other aromatics formaldehyde acetaldehyde propionaldehyde acrolein crotonaldehyde benzaldehyde other aldehydes glyoxals acetone methyl ethyl ketone other
gasoline engine exhaust non catalyst catalysta equipped* 9.5 0.9 0.1 1.3 0.2 1.9 3.7 0.8 1.3 1.1 0.3 0.6 0.6 2.2 0.2 2.8 1.4 1.8 1.4 0.3 0.2 1.2 0.4 0.5 0.2 8.3 10.2 4.3 0.8 1.0 0.2 0.5 0.6 1.5 1.6 g
2.3 1.1 1.1 0.2 0.1 1.2 0.6 3.6 5.8 1.2 1.2
1.6 4.2 0.5 0.3 0.6 1.8 1.3 2.30 0.63 0.10 0.47 0.12 0.41 0.21 0.14 0.04 2.2
unburned gasoline (summertime) headspace whole vaporsd liquid'
Emissions Composition' 15.0 2.9 0.1 4.4 0.5 1.8 4.2 1.0 2.1 1.3 0.6 1.0 0.3 3.0 0.2 7.6 1.2 1.1
0.7 0.0 0.2 0.5 0.0 0.4 0.3 2.6 5.9 2.5 0.4 1.3 0.6 0.3 0.4
0.9 0.3 0.5 0.1 0.1 0.2 0.1 0.1 0.2 0.3 3.7 8.9 1.3 0.9 1.3 3.5 1.9 0.4 1.3 1.9 0.5 1.7 1.20 0.62 0.06 0.14 0.03 0.20 0.31 0.39 0.08 2.0
3.3 0.8 2.7 6.9 2.0 3.3 2.1
0.2 1.1 2.0 8.8 1.0 9.4 4.9 1.2 0.2
0.2 2.1 30.0 11.4 6.3 22.3 1.1 2.8 1.6 0.3
0.2 0.4
0.1
0.0
0.8 0.1
0.1 0.1
0.4 0.5 3.7 0.5 1.1 0.3 0.1
0.1
1.3
g
g
0.1 0.1 0.4 2.4 0.3 2.0 0.0 0.8 1.0 0.2 0.5
9.6 0.9 0.2
1.1
0.2 1.8
0.0
0.5 2.5 0.6 1.0
commercial jet aircraft exhauste
1.4
4.2 17.4 5.2 2.0 k!
0.5
1.1
2.7 5.1 0.2 1.1 0.1 0.0 0.0
0.8 0.2
0.5 0.3 0.4
0.3 0.1 1.8
0.3 1.9 10.2 1.9 1.4 3.1 8.3 3.7 0.5 1.2 3.2 0.9
0.7 0.7 0.0
0.0 0.1 0.0
1.9 0.5 0.2 0.4 0.2 0.3
0.0 0.3 0.1
1.7 15.0 4.6 1.0 2.3 0.6 1.4 4.5 2.5 13.0
Exhaust speciation for gasoline-powered light-duty vehicles without catalytic converters. *Exhaust speciation for catalyst-equipped gasolinepowered light-duty vehicles. Composition of whole liquid gasoline. dComposition of gasoline vapors in headspace over liquid fuel. e Jet engine exhaust using landing and takeoff cycle times for commercial aircraft. 'Percent by weight of total organic gas emissions. #Compound not resolved separately from compound listed immediately above.
in the official emission inventory. The volatility of gasoline is seasonally adjusted, with increased vapor pressure during winter months to improve
cold weather starting characteristics and decreased vapor pressure in the summer to reduce evaporative emissions and to prevent vapor locking problems (15). This reducEnviron. Sci. Technol., Vol. 26, No. 12, 1992
2307
Table 111. Revised Speciation Profiles for Surface Coatings and Adhesives architectural coatings solventborne
chemical species
waterborne
thinning solvent
industrial coatings
industrial adhesives
1.0 3.1 1.7 0.6 1.3 3.2 0.5 0.5 20.7 11.6 5.9 10.3 16.0 0.6
1.8 5.7 3.1 1.1 2.4 5.8 0.9 0.9 14.7 15.8 4.9 3.1 8.1
11.4 10.8 5.8 2.1 4.6 11.0 1.7 1.6 5.6 0.1
11.8 5.0 0.3
2.7 3.5 6.4
Emissions Compositiona C9- alkanes C10 alkanes C11 alkanes Cl2+ alkanes C9- cycloalkanes C10 cycloalkanes C11 cycloalkanes Cl2+ cycloalkanes toluene xylene other aromatics acetone methyl ethyl ketone methyl isobutyl ketone methanol ethanol 2-propanol butanols ethylene glycol propylene glycol glycol ethersb ethyl acetate propyl acetate n-butyl acetate l,l,l-trichloroethane dichloromethane other miscellaneous
6.4 20.2 11.6 5.0 8.8 20.5 3.4 3.1 3.2 1.0 1.6 0.1 1.2 1.6 0.1 0.1 2.3 2.2 44 29 27
3.7 0.9 1.4
2.9 2.2 1.2
Percent by weight of total organic gas emissions.
Year fuel grade
1979
1987
unleaded regular unleaded premium leaded regular leaded premium
40 0 38 22
55 27 18 0
tion in fuel volatility has been achieved by reducing the amount of butane blended into summer gasolines (13,14). The 1987 composite summertime gasoline is 1%more reactive for whole gasoline and 12% more reactive for gasoline vapors per unit mass of emissions when compared to the wintertime composite fuel for the same year. Revised speciation profiles for summertime whole gasoline and gasoline headspace vapors are given in Table I1 and are used elsewhere throughout this study. 2.2. Gasoline Engine Exhaust Emissions. There are many published studies detailing the chemical speciation of vehicle exhaust for different vehicle types, driving conditions, and fuel types. These studies include dynamometer-based measurements for noncatalyst (16-19), oxidation catalyst (16,17,20-22),and three-way catalyst (23-25) equipped light-duty vehicles. Other measurements of exhaust speciation have been made in highway tunnels (26-30), in a parking garage (30),at the roadside (31),and on board a fleet of noncatalyst vehicles driven over a range of speeds (32). A number of important conclusions emerge from these studies. The exhaust from noncatalyst light-duty vehicles contains a higher percentage of acetylene and alkenes and a lower percentage of methane than is present in the exhaust from catalyst-equipped vehicles (16,20,28).Total emissions and detailed organic emissions vary with speed. A fleet of 46 vehicles tested on a dynamometer using three Envlron. Sci. Technol., Vol. 26, No. 12, 1992
5.9
6.6 2.0 1.6 3.1
0.2 0.1 0.4 0.2 0.1 0.2 0.2 0.3 9.3 1.7 18.2
* 2-Butoxyethanol, for example.
Table IV. Percentage Share of Total Gasoline Volume Sales by Grade in Los Angeles
2398
2.1 12.2
different driving cycles [a low-speed city driving cycle, the standard Federal Test Procedure (FTp) cycle, and a higher speed cycle representing conditions on a crowded urban expressway] showed decreased total emissions per mile driven and a higher methane fraction for the exhaust organic gas emissions as average speed increased (22). For noncatalyst vehicles tested on the road (32),total organic gas emissions decreased with increasing speed up to an average speed of 90 km/h and then began to increase as average vehicle speed was increased beyond this point. The fraction of combustion-derived species such as methane, acetylene, ethene, and propene in the total exhaust emissions increased with increasing speed. 2.2.1. Noncatalyst Vehicle Exhaust. The official inventory for Los Angeles speciates the exhaust emissions from noncatalyst light-duty vehicles on the basis of measurements made during the 1970s (20,33).Since that time, the allowable lead content of gasoline has been reduced, and gasoline suppliers have been forced to use other high-octane fuel components to make up for reductions in tetraethyllead use. The original exhaust speciation measurements did not include aldehydes as part of the exhaust emissions, so the official profile has weight percents of all species summing to 106% resulting from the addition of aldehydes to the list of emitted species. The level of chemical detail in the official profile is limited: only 24 species are listed, whereas more recent speciated exhaust measurements (19,W)have identified 100 or more species. In the course of evaluating the effect of a reformulated fuel on vehicle emissions, tests on older noncatalyst vehicles were performed by two independent laboratories (19). These tests were sponsored by the Atlantic Richfield Co. (ARCO) and used both a reformulated gasoline and a preexisting (1988) regular leaded gasoline. In the present study, a revised exhaust speciation profile for noncatalyst gasoline engines burning regular leaded gasoline was cre-
Table V. Summary of Selected Motor Vehicle Exhaust Speciation Experiments study name
ARCO Cohu et al. (19) vehicle sample size model years catalyst type gasoline grade leaded/unleaded fuel vapor pressure (RVP) (psi) octane number [(R + M)/2] fuel % alkanes fuel % alkenes fuel % aromatics fuel % unknown
2P 1970-1979 none regular leaded“ 8.9 88 60 10 30 0
EPA 46-car Sigsby et al. (22) 6b 1975-1982 variesd premium unleaded 8.8 92.9 46 9 44 1
40b 1975-1982 variesd regular unleaded 12.2 87.1 58 11 31 0
EPA low temp Stump et al. (23,241 20 1984-1987 3-way regular unleaded 11.5 62 5 30 3
auto/oil Burns et al. (25) 7e
1983-1985 mostly 3-way regular unleaded 8.7 87.3 59 9 32 0
The first 6 vehicles in the 46-car study were tested using premiuma Sample includes four light-duty trucks (1975-1979 model years). grade gasoline; the other vehicles were tested using regular-grade gasoline. Includes only the older (1983-1985) vehicles tested using industry average gasoline. dThe majority are oxidation catalysts. “Lead content is 0.08 g/gal.
ated by first averaging the weight percent speciation data for all vehicles tested at each of the two laboratories used in the ARCO study and then averaging, with equal weighting, the resulta reported by each laboratory to form the final speciation profile. The ARCO exhaust speciation data obtained using the regular leaded gasoline show significantly higher dialkylbenzene content than is indicated in the official speciation profile, while toluene content is lower. The ARCO test data indicate a lower butane content in the exhaust as well, possibly due to use of lowvolatility fuel prescribed during summer months in California. The official speciation profile for noncatalyst vehicle exhaust does not include any emissions of 1,3-butadiene, whereas the revised profile indicates 1.1% (by weight) of the total organic gas emissions for this species. The revised speciation profile for exhaust from older noncatalyst cars is 7% less reactive than the official speciation profile would indicate. 2.2.2. Catalyst-Equipped Vehicle Exhaust. The official emission inventory uses the FTP data from the EPA 46-car study (22)to speciate the exhaust emissions from catalyst-equipped gasoline-powered vehicles. More recent studies (23-25) have examined vehicles equipped with three-way catalyst systems. A summary of the vehicle fleets tested and fuels used is presented in Table V for selected exhaust speciation studies. Most of the vehicles in the &car study were teated using a regular-grade gasoline intended for wintertime use. This fuel is not representative of summertime fuels used in Los Angeles. Six vehicles from this study were tested using a premium summer-grade gasoline. Exhaust speciation data from these six vehicles have been averaged with data from another EPA study (23,241 and with data for older (1983-1985) vehicles from the auto/oil study (25)to form the revised speciation profile, with all vehicles within a single study averaged together to form composite profiles, and a final composite profile obtained by averaging results from each of these studies together, with equal weighting assigned to each study. The one-third weighting assigned to the speciation data for the 6 vehicles from the 46-car study tested using premium-grade gasoline matches the fraction of unleaded gasoline sales that was premium grade in Los Angeles in 1987 (24). Consideration of the use of premium-grade gasoline is important given the higher aromatic content typical of such fuels (see Table V). The reactivity of the revised speciation profile is 6% lower than that of the official profile. In the EPA 46-car study (221,benzene and cyclohexane peaks were not resolved separately during the gas chro-
matographic analysis. The authors estimated “that 50% of the amount stated is benzene”, but the more recent EPA (23,24)and auto/oil(25) studies indicate that essentially the whole sum should be counted as benzene, 80 the revised speciation profile counts all of the unresolved cyclohexane and benzene mass reported in the EPA 46-car study (22) as benzene. The revised speciation profile differs significantly from the official speciation profile in level of chemical detail and in the weight fractions of some individual species. In particular, the revised profile specifies less propane (0.1 % vs 2.2% in the official profile), more propene (2.5% vs 0.7%), and more benzene (3.7% vs 2.8%) than is specified in the official profile. While l,&butadiene is estimated in the revised profile to comprise 0.23% of exhaust emissions from catalyst-equipped vehicles, there are no lB-butadiene emissions specified in the official speciation profile. Emissions of 1,bbutadiene from catalpt-equipped vehicles are observed to occur mostly during the cold-start phase of the FTP test, because catalytic converters are very effective in removing l,&butadiene once they warm up (23, 24, 34, 35). A recent old-car buy-back program in the Los Angeles area provided exhaust emission rate measurements for a fleet of pre-1971 cars (36).These old noncatalyst cars were reported to emit on average 16.5 g of exhaust hydrocarbons per mile traveled, representing an average exhaust mass emission rate 66 times higher than new (1990) cars equipped with modern catalytic converters (36). The speciation of emissions from the vehicles in this old car buy-back program is not well understood, but if it should turn out that their exhaust composition is similar to that of the other noncatalyst cars tested previously, then these old noncatalyst vehicles may emit over 300 times the masa of l,&butadiene per mile driven when compared to new catalyst-equipped vehicles. As shown in Table I, there were comparable NMOG mass emissions from noncatalyst and catalyst-equipped vehicles in Los Angeles for 1987. Therefore the noncatalyst vehicles likely contributed the majority of on-road vehicle emissions of 1,3-butadiene, as can be seen by comparing the speciation profiles given in Table I1 for catalyst-equipped and noncatalyst vehicle exhaust. 2.3. Diesel Engine Exhaust. The official inventory uses a speciation profile for diesel engine exhaust based on Table 9-07-021 of the EPA species data manual (37), with additional aldehyde speciation data. The compound n-pentadecane is used as a surrogate for a range of diesel fuel constituents present in the exhaust. The detailed Environ. Sci. Technol., Vol. 26, No. 12, 1992 2399
composition of the fuel-derived exhaust emissions is expected to depend strongly on the characteristics of the crude oil from which the diesel fuel was refined (38). In the revised inventory, we have retained the existing profile for lack of more detailed speciation data. A second diesel exhaust profile that was used in parts of the official inventory indicates a sum of weight percents for all species greater than loo%, following an assumption that the inventory did not include aldehydes in the total mass emission rates. Since all emission rates are now specified in terms of total organic gas rather than total hydrocarbon mass, all diesel exhaust organic gas emissions are now speciated using the first profile described above. 2.4. Jet Engine Exhaust. The organic gas emissions from aircraft jet engines were investigated by Spicer (39). Based on these tests and landing and takeoff cycle times for commercial and military aircraft, new and separate speciation profiles for commercial and military jet aircraft exhaust (2)have been used in place of the speciation profile defined in the official emission inventory. The official speciation profile lists only 10 species in jet engine exhaust, whereas the revised profile includes over 50 species. These changes result in a 23% increase in the reactivity of jet engine exhaust emissions. 2.5. Surface Coatings. The emission inventory for Los Angeles contains significant organic gas emissions attributed to the use of surface coatings, as shown in Table I. The speciation profiles used in the official emission inventory are generally based on limited analyses of surfacecoating samples and fail to resolve significant fractions of the sample mass in some cases. Our approach in developing revised speciation profiles for organic gas emissions from surface coatings has been to use manufacturer-supplied data indicating overall usage of organic solvents in various classes of surface-coating products. Petroleum distillates (also referred to as mineral spirits) are a common ingredient in many surface-coating products. Analyses of the detailed chemical composition of typical petroleumbased solvents were available for several different solvent suppliers (40),and these analyses show that a range of C8-CI2 alkanes and cycloalkanes are the main chemical constituents of these solvents. This information has been incorporated in the revised speciation profiles for organic gas emissions from surface coatings. 2.5.1. Industrial Surface Coatings and Adhesives. In the official emission inventory, a composite speciation profile developed by Oliver and Peoples (13)is used to speciate emissions from industrial surface coating activities. This profile is based on direct chemical analyses of composite samples of primers, lacquers, and enamels (13).In the present study, a profile presented in Table 6.1-1 of Rogozen et al. (41)has been adopted. This profile is based on solvent consumption by the manufacturers of industrial surface coatings. The respeciated emissions are 37% more reactive than is indicated by the official inventory. Further study of the chemical composition of such coatings is still needed. Solvents incorporated in industrial adhesives in California are listed in Table 6.5-1 of Rogozen et al. (41).This listing has been used instead of the existing adhesive speciation profile (13)and results in a 3% increase in reactivity. The official speciation profile lists 56% of the organic gas emissions as isomers of pentane, while the revised profile is dominated by mineral spirits and naphtha (48%), methyl ethyl ketone (15%), and l,l,l-trichloroethane (11%). 2.52. Architectural Surface Coatings. Architectural surface coatings generally can be classified as either organic 2400
Envlron. Scl. Technol., Vol. 26, No. 12, 1992
solvent- or water-borne. The official emission inventory for Los Angeles indicates that there are substantial organic gas emissions due to application of both types of coatings (see Table I). A recent survey by the California Air Resources Board (42)reported the sales volumes, estimated VOC content, and associated thinning solvent usage for 38 different categories of solvent-borne architectural surface coatings. Similar data are reported for water-borne coatings. In the present study, the average organic solvent composition for the major coating categories has been determined from a review of manufacturer-supplied material safety data sheet (MSDS) information (43-47). The composite solventborne coating speciation profile was developed by averaging the results obtained for each of the major coating categories using the estimated VOC emissions for each coating category (42)as weighting factors. Petroleumbased solvents such as mineral spirits, varnish makers, and painter’s naphtha comprise about 70% of the volatile organic ingredients used in solvent-borne architectural coatings. One solvent-borne coating category, lacquer, contains a completely different set of chemical compounds. The detailed chemical composition of lacquers was determined from manufacturer-supplied MSDS information, with the major organic ingredients being acetates, alcohols, and ketones. The official speciation profile (from Table 9-35-103in ref 37) uses n-hexane and cyclohexane as surrogates for all of the alkanes contained in mineral spirits. As described previously, minerals spirits contain a range of C8-Cl2 alkanes and cycloalkanes (40),and this is now reflected in the revised speciation profile. The toluene and xylene content of the revised speciation profile is reduced. The respeciated organic gas emissions from use of solvenbborne architectural coatings are 24% more reactive than the official inventory would suggest. There are additional organic gas emissions associated with the use of solvent-borne coatings due to cleanup and thinning operations. Chemical analysis of a composite thinning solvent sample (13)provides the basis for speciating these emissions in the official inventory. However, that analysis left 41% of the sample mass unidentified. The composition data for thinners presented in Table 6.1-1 of Rogozen et al. (41)have been used to respeciate these emissions; the reactivity of thinning solvent emissions is found to decrease by 61% after these changes are made. According to Rogozen et al. (41)and MSDS information, the principal organic cosolvents in water-borne coatings are ethylene glycol, propylene glycol, and a variety of glycol ethers and esters. This list of compounds does not correspond to the major species identified in the existing speciation profile (13).It may be that low-volatility compounds such as the glycols were not recovered during the distillation step in the analysis of paint samples that is used in the official inventory. A revised speciation profile for water-borne coatings has been developed based on data from Table 6.1-1 of Rogozen et al. (41)and is 117% more reactive than the profile used in the official inventory. 2.6. Other Changes. In the official inventory for Los Angeles, some source categories have been assigned to a generic “species unknown” profile. In some cases, an alternative profile exists that is better related to the actual source category. Organic gas emissions that occur during cleaning and priming of vehicles prior to repainting and emissions from other miscellaneous surface-coating activities have been reassigned to the composite industrial surface coating profile described previously. Organic solvent usage for thinning and cleanup operations asso-
camporite inven,ory Landfill emisrionr
Gas distribution
j
0
C10 alkanes C t l alkanes glycol ether
OlliCiSi
]
A-!,
Adhesive3
lndurlrial coatings Water-borne pain, Sal"e"f-bornc
pa,",
Jet engine e i h s u l
Gssoiine vapors Whole gasoline
Cafslyrt ven\c,e exhsYO, Non.cafslysf
,
,
~
C
I I
00
.
'
,,
I_jl
~~~~
0 5
I 0
1
5
2.0
2.5
3.0
Reacflvity Index
Flgun 1. Comparison of hydroxyl radical reactkty f w organic gas emlssion speciation profibs in the official and revised emission Inventories.
ciated with use of architectural surface coatings was also assigned to the species unknown profile in the official inventory. In our revised emission inventory, these organic gas emissions are reassigned to the composite thinning solvent profile described earlier. These changes affect 50 tons of organic gas emissions per day. After reassignment of source categories to alternate speciation profiles, only 49 tons per day of the nonmethane organic gas emissions originally assigned to the species unknown profile remains in that category. A speciation profile for the organic gas emissions from commercial cooking operations would help to reduce this total still further: presently the official inventory includes 19 tons per day of organic gas emissions from commercial charbroiling, deep fat frying, and other unspecified operations (excluding baking). Running evaporative emissions from motor vehicles were mislabeled as crankcase emissions and assigned to the noncatalyst exhaust speciation profile in the official emission inventory. This source category (56 tons of organic gas emissions per day) was reaasigned to the gasoline vapor speciation profile instead. Organic gas emissions from use of aerosol sprays include both propellant and solvent emissions. The propellanta are typically light hydrocarbons such as propane and butane and synthetic compounds such as dichloromethane. The solvent emissions were speciated as if they were additional propellant. In the revised inventory, the solvent emissions (41 tons/day) have been reassigned to an existing composite domestic solvent speciation profile. Composite degreasing and cold cleaning solvent speciation profiles were derived from the official emission inventory using all emision data where the actual degreasing solvent being used was specified. These composite profdes were used to speciate miscellaneous degreasing and cold cleaning emissionswhere the actual solvent was not known. The old composite profile in the emission inventory referred to solvents such as trichloroethylenewhich are b e i phased out. Evaporative emissions from crude oil production facilities were reassigned from a profile which specified emissions in terms of carbon bond mechanism species (48)to another crude oil evaporation profile that gives explicit chemical speciation. 2.7. Summary of Revisions. The changes in the reactivity of key speciation profiles that occur when the revised speciation data are introduced are summarized in Figure 1. The revised speciation profiles have important implications for the evaluation of emission control pro-
i
ethylene glycol propylene glycol 1.3-buladiene benzene formaldehyde acetaldehyde butene methanol propene ethene n-hexane cyclohexane
p--
B B
t-
,,
.-
exhaust
I
C12 alkanes
7 1
Thinning ~ ~ I v e n f ~
t
I PA
n
-60 Emissions Difference (Ionstday) (rerpeeiated minus official inventory)
F l p a 2. change (tonsldsy) in bashwlde emissbm after reapec+atbn fw selected wmpounds.
propylene glycol C t t alkanes
ethylene glycol 1.3-butadiene 01
1
1
10
100
Emission Ratio (ReopeciatedlOlficiaI Inventory)
*re
3. Percentage of baslnwlde emlsslons in the respeclated vs
officialem)ssion Inventory lw sekcted compounds.
posals because the relative importance of various contributors to the total inventory is now represented better.
The largest changes in reactivity for individual Speciation profiles were observed for surface-coating and solvene usage categories. The reactivity of the overall organic gas emission inventory increases by 4% after respeciation. While the preceding discussion has focused on differences in reactivity between official and revised speciation profiles, the main motivation for the respeciation effort conducted in the present study is to advance the ability to compare emission data to ambient air quality for single organic compounds. The adoption of revised speciation profdes has led to major changes in the emission estimate for many individual chemical compounds. Some of the more striking differences are shown in Figures 2 and 3. Respeciation of the organic gas emissions results in 14 times greater emission estimates for lB-butadiene, a toxic air contaminant. Emissions of cyclohexane are reduced to 9% of the official inventory estimate. Use of new speciation profiles has also improved the level of chemical resolution in the inventory. In some eases, emissions of various isomers of a single compound such as hexane or xylene were formerly lumped together as a single sum. The respeciated inventory is more precise in specifying which isomers are emitted. Envlron. Sei. Technoi.. Voi. 26. No. 12. 1992 2401
1000
1"
t
l
0
0
0
1000
1000
1
C c
c
E w a,
9 I
I
m c
2 C
pi
0
C 0
O
T
1 10
u
c
.-na,
.-C m
0 .-c
0
.-
0
T
E U
c - ;. ; -t P i&
Figure 4. Abundance of alkanes in the basinwide emission inventory and in ambient air at Giendora, CA.
3. Comparison with Ambient Data During the period August 10-21, 1986, ambient concentrations of gas-phase organic species were measured downwind of Los Angeles in Glendora, CA, as part of the Carbonaceous Species Methods Comparison Study (8). Samples were collected over 4- or 8-h intervals covering all 24 h for each day of the study. Average concentrations for each organic species have been calculated using data from all valid sampling periods and all days. Several of the samples were deleted because of problems with flow rate regulation during filling of the stainless steel sample canisters (49). Comparisons between the revised basinwide emission rates for individual compounds and the observed ambient concentrations are presented for selected alkanes in Figure 4 and for alkenes, aromatics, and chlorinated compounds in Figure 5. In order to reflect only the methane contributed by local emission sources, the average ambient concentration plotted for methane in Figure 4 is reduced below the measured value by 1700 ppb, a value representative of upwind background methane levels at San Nicolas Island (50). The selection of species included in these figures was limited to those that are available in the ambient data base; there are about 200 additional compounds specified in the emission inventory. The relative positioning of the vertical axes for emission and ambient concentrations in these figures is arbitrary. The fact that the displacement of ambient concentration points from plotted emission points for the corresponding species is fairly uniform indicates that the relative emission rates of the various species in the revised emission inventory are consistent in order of magnitude with the atmospheric data base. For the alkanes shown in Figure 4, the respeciated inventory generally follows the same trends as the ambient data in terms of the relative abundance of species. Propane, methylcyclopentane, and methylcyclohexane emissions are likely understated relative to emissions of other alkanes, whereas n-nonane emissions appear to be overstated. In Figure 5, similar comparisons are presented for alkenes, acetylene, aromatics, and several chlorinated species. The relative abundance of these compounds in the res2402
1
C
0
n
._0
Emissions
Environ. Sci. Technol., Vol. 26, No. 12, 1992
r
.-
Figure 5. Abundance of alkenes, aromatics, and chlorinated species in the basinwide emission inventory and in ambient air at Giendora, CA.
peciated organic gas inventory generally agrees with the ambient data. As shown in Figure 5, the actual emissions of ethylbenzene, ethyltoluene, trichloroethylene, and perchloroethylene would appear to be higher than stated in the emission inventory, based on the closer than usual spacing of the emission and ambient data points for those compounds. The 2-butenes are highly reactive so one should expect that the ambient data point wiU be displaced further below the emission point relative to the typical displacement for other species, as seen in Figure 5. 4. Source-Receptor Reconciliation
It is highly desirable to have independent methods for checking the accuracy of the emission inventory before using it in photochemical modeling calculations. The chemical mass balance technique (51,52)can be used to determine the contributions of various pollution sources to ambient pollutant concentrations at receptor sites where air quality measurements are available. By using the chemical mass balance model to examine source contributions at many receptors, it is possible to determine whether the emission inventory is accurate in its description of the relative importance of the various source types. Such source-receptur reconciliation techniques have been used before for Los Angeles (53),Sydney (54),New Jersey (55),and Chicago (56,57). Since chemical mass balance models do not directly employ the mass emission rate data, such an analysis does not determine the absolute magnitude of the source emission rates in tons per day-only the relative importance of the source types is revealed. As demonstrated by Cass and McRae (58),the tracer species used for receptor modeling must be selected carefully. The emission inventory for each chemical species should be examined to determine which source categories are thought to be significant contributors to the emissions of each chemical species under consideration. A compound should not be used in the model if the set of source speciation profiles available for use in the receptor model does not include all significant sources of the compound. In chemical mass balance models, the mass concentration of a species at a receptor air monitoring site is taken
3800 Burbank 0
3780.
Glendora oclaremont ~Rubidoux *Anaheim
3740. 3720
-
PACIFIC OCEAN
3700 3680 200
w a n Nicolas Island
240
280
320
360
400
440
480
520
UTM Easting (km)
Figure 6. Map showing receptor monitoring sites where speciated organic gas concentration measurements were made.
to be a linear combination of the emissions of that species from various sources: (4)
where c i k is the mass concentration of species i at receptor k , aij is the weight fraction of species i in the total direct emissions from source j , fi,k is the fractionation coefficient of species i from source j , and s j k is the total mass contribution of all species emitted from source j to air quality observed at receptor k. There are m different source categories, and the unknowns to be determined are the relative source contributions s j k . The fractionation coefficient, f i j k , is the fraction of species i from source j which will reach receptor k. This coefficient is used to account for loss of species mass during transport from source to receptor due to removal processes such as atmospheric chemical reactions and deposition at the earth’s surface. For this analysis the organic gas tracer species were selected to exclude highly reactive species so that the fractionation coefficients can be set to unity. For a single receptor k,eq 4 is written repeatedly for each of n chemical species. This set of equations forms an overdetermined system provided that the number of species is greater than the number of source categories to be resolved. The system is solved by regression analysis to obtain the best fit combination of source contributions that reproduces the ambient concentrations of all species with minimum squared error. In this study, the CMB7 source-receptor modeling computer program (59) has been used to solve the set of equations (4) at each receptor k. This program requires ambient concentration data ( c ~ and ) source speciation data (aij)as inputs. Uncertainties to be associated with each data point are also required. The standard deviation of the determination of the weight percent abundance of each species in the ambient data set is used to weight Cik and a i j so that each species is treated relative to the precision with which it can be measured. Without such weighting, species with large ambient concentrations would dominate the calculation, and valuable concentration data obtained for trace species would be ignored. Confidence intervals on the source contribution estimates are computed by the CMB7 program using speciation profile and ambient concentration uncertainty data. 4.1. Source, Receptor, and Species Selection. The monitoring sites used in the receptor modeling calculations are shown in Figure 6. The list of sites includes Glendora [special monitoring site for the Carbonaceous Species
Methods Comparison Study (81,August 10-21, 19861, Claremont, Long Beach, Anaheim, Azusa, Burbank, central Los Angeles, Hawthorne, and Rubidoux [intensive monitoring sites during the Southern California Air Quality Study (SCAQS) (9), summer and fall 19871. Unlike the samples collected at Glendora which were 4- and 8-h-average samples, the hydrocarbon samples collected during SCAQS were obtained over l-h sampling intervals. The samples were collected in stainless steel canisters by Aerovironment (60) at 0600,1100, and 1500 PST, with additional samples collected at the Claremont and Long Beach sites only at 0400, 0800 and 1300 h PST. The samples were analyzed using gas chromatography as described by Stockburger et al. (61) and Rasmussen (62). Average ambient concentrations were determined for each species using all available summertime data at each receptor. The corresponding uncertainties in the ambient concentration data for each species were computed from the standard deviation of the weight percent of each species on a sample by sample basis. This was done to normalize for meteorological variations that affect the absolute species concentrations but not necessarily the relative abundances. The number of source categories that can be resolved is limited by availability of tracer species in the ambient data seta and by collinearity among the speciation profiles being used. There are over 800 individual source categories referenced in the emission inventory and only about 75 species present in the ambient data base that could be used to resolve separately all of these sources. Fortunately, a relatively small number of sources contribute most of the organic gas emissions (see Table I). I t is still necessary to lump some of the speciation profiles for different source categories together when there are no appropriate tracer species in the ambient data set for a source category or when speciation profiles are nearly collinear. Further it is necessary to screen the species used to assure that they react slowly enough in the atmosphere to justify the approximation that f i j k is unity, and also to assure that the few major sources included in the calculation contribute the overwhelming majority of emissions of those compounds such that the material balance of eq 4 is in fact observed. The lumping and screening procedures employed here follow the emission inventory assisted receptor-modeling methods of Cass and McRae (58). Source categories with similar speciation profiles and similar spatial distribution are pooled. The lumping was carried out using total organic gas emissions (as per the revised emission inventory) to weight the individual speciation profiles to form a composite profile for each lumped group Environ. Scl. Technol., Vol. 28, No. 12, 1992
2403
Table VI. Source Speciation Profiles for Use i n Receptor Modeling
compound" methane ethane ethene acetylene n-butane n-pentane iS0pe"tane n-hexane branched C6 alkanes methylcyclopentane methylcyclohexane benzene ethylbenzene perchloroethylene l,l,l-trichloroethane all other species
engine exhaust
waste and natural gas
gasoline headspace vapors
dry cleaning solvents
whole gasoline
degreasing solvents
Emissions Compositionb 84.5 f 15.9 4.6 f 0.5 0.2 f 0.1
11.9 f 4.0 1.9 f 0.6 8.2 f 2.0 5.3 f 2.0 2.9 f 0.8 1.9 0.4 3.7 1.2 1.0 0.2 4.1 i 0.6 0.8 f 0.2 0.4 f 0.2 3.6 f 0.7 1.2 f 0.2
2.1 0.3 0.9 0.2 1.1 0.2 0.5 0.1 0.6 f 0.1
*
0.2 f 0.1
* *
3.3 f 0.6 2.7 f 1.4 6.9 f 1.9 2.0 f 0.9 6.7 1.5 2.5 1.2 1.0 0.5 1.9 0.1 1.9 0.3
30.0 6.3 6.3 f 3.0 22.3 4.9 1.1 0.3 5.7 1.0 1.1 0.4 0.1 f 0.1 0.7 f 0.1
* * * *
94 f 10
6
53
32
71
6
*
11.5 2.7 43.5 f 9.3 45
'Includes only those species which are present in the ambient data base and meet the other screening criteria described in the text. bPercent hy weight of total organic gas emissions for the specified lumped source category.
Engine Exhaust Whole Gasoline
Gasoline V a p o r s Waste 8 Natural Gas Degreesing Solvents
Other
0
20
40
60
80
100
Percent of Basin-wide Emss!ons
Flguro 7,
Percentage of baslnwlde emissbns of individual specles atmbuted to various SOurcx categwles
of sources. The lumped sources and their composite speciation profiles are shown in Table VI. The standard deviations in Table VI were estimated from speciation uncertainties for each source category and from the uncertainties associated with the lumping procedure. The lumped sources shown in Table VI account for approximately 60% of the anthmpegenicnon-methane organic gas emissions in the South Coast air basin. The source category labeled waste and natural gas includes emissions from methane-rich sources such as natural gas seeps and distribution losses, waste decomposition, oil and gas production, and certain petroleum refinery emissions. A screening procedure was applied to each candidate tracer species for use in the receptor modeling calculation. The first criterion was that measurements of ambient concentrations for the species must be available. In addition, the species must not be too reactive: assuming a hydroxyl radical concentration of lo6 molecules/cm3, the species must have a half-life with respect to hydroxyl radical attack of a t least 20 h. Peak hydroxyl radical concentrations in heavily polluted areas such as Los Angeles may approach lo7molecules/cm3 at midday during the summer, but will be much lower at other times of day (63). Furthermore, the species should not be formed in 2404
Envlron. Scl. Technol., Vol. 26. No. 12. 1992
appreciable amounts from the atmospheric oxidation of other organic species. Finally, at least 75% (typically more than 80%) of the emissions for the species must be accounted for in the source speciation profiles being used. Only those measured species which meet the above criteria appear in Table VI. Initially, all species shown in Figure 7 were considered for inclusion in the receptor-modeling calculation. Due to the following considerations,some of those compounds and the source classes that they might trace have been eliminated. The fraction of perchloroethylene emissions attributed to 'other" sources in Figure 7 results almost entirely from dry cleaning solvent use. Perchloroethylene thus would be an excellent candidate for a fitting compound within our analysis and could be used to sort out emigsions from dry cleaners. However, perchloroethylene concentration data were only available for the Glendora monitoring site. Therefore, at this time, perchloroethylene data cannot be used consistently throughout our analysis. Significant fractions of the emissions of toluene, xylene, and acetone also are attributed to source8 other than those listed. Acetone would be a useful tracer for domestic solvent use, except that it is formed in the atmosphere as an oxidation product of other organic gases and, therefore,
Table VII. Sonree Contributions' to Non-Methane Organic Gas Concentrations
receptor
G1endora
Anahnim .I.-._.I
hUSa
central LA Claremont Hawthorne Long Beach Rubidaur ~~
engine exhaust
waste and natural gas
gasoline headspace vapors
whole gasoline
34 11.4 + 29 __. - _159 f 47
Source Contribution (pg/mS) 46*9 21 20 177 33 48 20 120 . .f 11 -~ 45 f 11 29 f 32 160
911
Ea
152
+dl
208 f 51 143 f 35 70 f 24 105 36 119 29
* *
* 46 36 85
17
41 10 25 7 44*10 49 11
32 t 17 42 20 27 19
*
107 f 45 124 38
*
dry
cleaning solvents
degreasing solvents
18 f 13 NDb ND ND
49 15 39 43 37 26 14 40 17
ND ND
23
*6 * 21
*
16 25 13 10 65 11
"Source contributions to nan-methane organic gas concentrstions at each receptor are reported with associated standard errom (*one standard deviation). b N o t determined, as the perchloroethylene concentration data needed were not available at these sites.
could not be used in the receptor-modeling calculation. The domestic solvent speciation profile also includes alcohols and glycols for which no ambient concentration data were available. Surface-coating activities emit a complex mixture of organic gases, as shown in Table 111. Of these compounds, only toluene and xylene are consistently included in the ambient concentration data base. Xylene does not meet the reactivity criterion specified above, and toluene by itself is not sufficient to act as a tracer for surface-coating activity, so the surface-coating source category could not be included in the final analysis. As a result, toluene was not used as a fittiig compound in the receptor-modeling calculations. Having explained why we will not attempt to calculate surface-coating and domestic solvent contributions to the ambient organics burden, it is now important to explain why that decision will not result in misidentification of solvent vapors as if they were unburned gasoline emissions. While it is true that certain species are present in both unhumed gasoline and surface-coatingsolvents, steps have been taken to guard against such a miscalculation. The main applications of the petroleum distillate solvents are for surface coating, degreasing, and cleaning. In surface coatings and in Stoddard solvent used for degreasing and cleaning, the bulk of the species are in the C W 1 2 alkane range and these compounds are not used as fitting compounds in the model. In the case of toluene and xylene, which are prominent both in gasoline and in some solvents, again the compounds are disquaMed as fitting compounds in the model. Solvent emissions from dry cleaning and degreasing can be tracked using chlorinated compounds such as perchloroethylene and l,l,l-trichloroethane when ambient data are available. The hydrocarbon solvents used by those source categories are carried along with the chlorinated compound tracer analysis, but these hydrocarbons are not used as tracer compounds in a way that could interfeze with the identification of unbumed gasoline in the atmosphere. The performance of the chemical mass balance model can be improved when tracer species are available that are unique to a single source category. Carbon monoxide, though not strictly speaking an organic gas, is a very desirable species to include in the analysis because it has a long atmospheric lifetime, is emitted almost entirely from motor vehicle exhaust, and was measured along with gas-phase organic species by gas chromatographic methods during the special field studies. The mass ratio of carbon monoxide to organic gas emissions for mobile sources used here is 8.7 i 3.1, based on measurements made in a Los Angeles area roadway tunnel (3). This value is consistent with ratios computed for other roadway tunnel experiments (28,64).
0
E 4
Flgum 8. Relative w r c e conblbutbns to nOrrmemane w n l c gas concentrations at various r e ~ e pmonkwing t~ skes.
Other processes that remove hydrocarbons from the atmosphere have been considered before the final selections of tracer species were made. Dry deposition is the only pnxzas other than atmospheric chemical reaction that removes various species from the atmosphere at differing rates. Because species that exhibit high removal rates via dry depition are highly reactive or highly soluble in water (65),and the hydrocarbons chosen as tracer compounds are not very reactive and typically have low aqueous solubilities, dry deposition is not likely to be a significant removal process for our tracer species over the time scales necessary for the present receptor modeling study. 42. SoumReceptor Modeling Results. The resulta of the chemical mm balance model analysis are presented in Table VI1 and in Figure 8. Table VI1 specifies source contributions to the non-methane organic gas concentrations observed at each receptor. A direct comparison b e tween the sum of computed source Contributions and observed organic gas concentrations to determine the residual contribution from other sources is not poasible because the emission inventory includes approximately 300 tons of oxygenated species per day such as alcohols and glycols that were not measured in the ambient data base, and because the most reactive compounds (not used as fitting compounds) will be depleted significantly in ambient air relative to the amount emitted. In Figure 8, the relative contributions of five lumped sources (engine exhaust, waste and natural gas, whole gasoline, gasoline headspace vapors, and degreasing solvents) at each receptor are compared with the relative importance of these sources as specified in the basinwide emission inventory. Note that a significant fraction of the basinwide organic gas emissions, including biogenic Envlron. Sd.Technol., Vol. 26, NO. 12. 1992 2405
emissions, domestic solvent use, and architectural and industrial surface coating activities, is not shown in Figure 8, and therefore, the relative source contributions are relative only to the sum of the five sources mentioned, not the total inventory. Inspection of Figure 8 reveals that the relative source contributions are fairly similar at all of the receptors. When compared to the relative source contributions specified in the emission inventory, the resulta at all of the receptors indicate that the emission inventory understates the emissions of whole (unburned) gasoline relative to exhaust emissions. The receptor model results do not, however, pinpoint the source of such excess unburned gasoline emissions, and there are many possible sources including understated evaporative emissions (e.g., additional hot-soak emissions or fuel spillage) and more unburned gasoline in tailpipe exhaust than is suggested by FTP-based testing. Possible mechanisms that would lead to additional emissions of unburned fuel in tailpipe exhaust include difficult cold starts and misfiring engine cylinders. Excess gasoline emissions might also occur during off-cycle conditions such as rapid decelerations that are not represented in F"P-based tests. A recent remote surveillance study has confirmed that a small fraction of the vehicle fleet contributes a disproportionately large share of the total exhaust emissions (5). Such vehicles are also important in determining the overall speciation of the exhaust emissions, but measurements made to date have not investigated the chemical speciation of exhaust emissions from the highest emitting vehicles. Further study of the many potential sources of the excess emissions described above is recommended. The results of Figure 8 are expressed in terms of percentage contributions to the ambient concentrations and do not explicitly determine the actual tonnage of gasoline vapors and exhaust emitted. However, if one were to assume that the exhaust emissions in the official emission inventory at least are not overstated, then the receptor modeling results suggest that the absolute emissions of unburned gasoline to the atmosphere are much greater than previously thought. For example, if the mass of non-methane organic gas emissions assigned to gasoline and diesel engine exhaust in the respeciated emission inventory (550 tonsfday) were correct, then the approximate one to one ratio of unburned whole gasoline to FTP-determined tailpipe exhaust found in the atmosphere suggests an emission rate of unburned gasoline of about 550 tonsfday compared to only 108 tonsfday in the current emission inventory. Since the exhaust emissions themselves may be understated (4-7), the actual mass emission rate of unburned gasoline could be even greater than the value just calculated. Resolution of this issue is complicated because the newly recognized emissions of unburned gasoline could be emitted in the tailpipe exhaust under conditions not represented by conventional FTP exhaust speciation tests. Such conventional FTP testa have already identified a 50% contribution of unburned fuel components to engine-out and tailpipe exhaust for three catalyst-equipped vehicles (35). The results of the present study suggest that this contribution might be even higher for the vehicle fleet and driving patterns typical of Los Angeles in 1987. 5. Conclusions Respeciation of the organic gas emission inventory has resulted in large changes in the basinwide emission estimates for many important organic gases. The reactivity of individual speciation profiles has changed significantly. The reactivity of the overall organic gas emission inventory 2406
Environ. Scl. Technol., Vol. 26, No. 12, 1992
showed only a modest increase of about 4% after respeciation, which may explain why photochemical models for ozone production have worked reasonably well in spite of major errors in the composition of the actual emissions. The relative abundance of individual species and species groupings in the respeciated emission inventory corresponds to the patterns seen in ambient concentration measurements taken at Glendora during August 1986. Receptor-modeling calculations indicate that the current emission inventory for the Los Angeles area understates the emissions of unburned gasoline relative to combustion-derived hydrocarbons such as ethene and acetylene. Excess unburned fuel may be emitted in tailpipe exhaust under driving conditions not represented in current ETP-based tests or if the vehicles used in speciated exhaust measurement studies are not truly representative of the on-road vehicle fleet. As well, evaporative emissions including hot soaks and fuel spillage may be underestimated relative to exhaust emissions in the official emission inventory. It was not possible to determine source contributions in these receptor-modeling calculations for biogenic hydrocarbon emissions (due to their short atmospheric lifetime) or for surface-coatingand domestic solvent use (due to lack of ambient concentration data for key marker compounds). There is a need for further study of the chemical composition of industrial surface coatings and the detailed composition of petroleum distillate solvents incorporated in surface coatings. In the future, it is recommended that speciated organic gas measurement programs include alcohols (especially methanol, ethanol, 2-propanol, and butanols), glycols (ethylene glycol and propylene glycol), and glycol ethers (e.g., ethylene glycol monobutyl ether) as part of the ambient concentration data sets. Availability of such data for these species would help to resolve the contributions from organic gas emission sources such as surface coatings and domestic solvent use. In addition to the alcohols and glycols, other species such as methyl tert-butyl ether should be measured as they are incorporated in reformulated motor vehicle fuels. Acknowledgments The authors gratefully acknowledge the assistance of Paul Allen and Doug Lawson of the California Air Resources Board in supplying supplemental data. The authors also thank Larry Rapp of ARC0 and Fred Stump of EPA for supplying vehicle exhaust speciation data, and Robert Wendoll of Dunn-Edwards Co. for helpful discussions and supplemental data relating to architectural surface coatings. Literature Cited Air quality management plan for the South Coast Air Basin. South Coast Air Quality Management District, El Monte, CA, 1991.
VOCIPM speciation data system documentation and user's guide, version 1 . 3 2 ~EPA-45012-91-002; ; Radian Corp.: Research Triangle Park, NC, 1990. Ingalls, M. N.; Smith, L. R.; Kirksey, R. E. Measurement of on-road vehicle emission factors in the California South Coast Air Basin. Volume I: regulated emissions. Report to the Coordinating Research Council under Project SCAQS-1, Southwest Research Institute,San Antonio, TX, 1989. Pierson, W. R.; Gertler, A. W.; Bradow, R. L. J . Air Waste Manage. Assoc. 1990,40, 1495-1504. Lawson, D. R.; Groblicki, P. J.; Stedman, D. H.; Bishop, G. A.; Guenther, P. L. J . Air Waste Manage. Assoc. 1990, 40, 1096-1105.
(6) Carlock, M. A. Mobile Source Division, California Air Resources Board, El Monte, CA, personal communication, 1992. (7) Fujita, E. M.; Croes, B. E.; Bennett, C. L.; Lawson, D. R.; Lurmann, F. W.; Main, H. H. J. Air Waste Manage. Assoc. 1992,42, 264-276. (8) Lawson, D. R.; Hering, S. V. Aerosol Sei. Technol. 1990, 12, 1-2. (9) Lawson, D. R. J. Air Waste Manage. Assoc. 1990, 40, 156-165. (10) Wagner, K. K.; Allen, P. D. SCAQS emissions inventory for August 27-29,1987 (Tape ARA714). Technical Support Division, California Air Resources Board, Sacramento, CA, personal communication, 1990. (11) Carter, W. P. L. Atmos. Enuiron. 1990,24A, 481-518. (12) Carter, W. P. L.; Atkinson, R. Enuiron. Sci. Technol. 1989, 23,864-880. (13) Oliver, W. R.; Peoples, S. H. Improvement of the emission inventory for reactive organic gases and oxides of nitrogen in the South Coast Air Basin. Report to the California Air Resources Board under Contract A2-076-32, Systems Applications, Inc., San Rafael, CA, 1985. (14) Kulakowski, J. M. UNOCAL Refining and Marketing, Los Angeles, CA, personal communication, 1990. (15) Gary, J. H.; Handwerk, G. E. Petroleum refiningtechnology and economics, 2nd ed.; Marcel Dekker, Inc.: New York, 1984; pp 8-9. (16) Jackson, M. W. S A E Tech. Pap. Ser. 1978, No. 780624. (17) Black, F. M.; High, L. E. J. Air Pollut. Control Assoc. 1980, 30,1216-1221. (18) Nelson, P. F.; Quigley, S. M. Atmos. Enuiron. 1984, 18, 79-87. (19) Cohu, L. K.; Rapp, L. A.; Segal, J. S. EC-1 emission control gasoline. ARC0 Products Co., Anaheim, CA, 1989. (20) Black, F. M.; High, L. E. SAE Tech. Pap. Ser. 1977, No. 770144. (21) Smith, L. R. Characterization of exhaust emissions from high mileage catalyst-equipped automobiles; EPA-4601 3-81-024; Southwest Research Institute: San Antonio, TX, 1981. (22) Sigsby, J. E.; Tejada, S.; Ray, W.; Lang, J. M.; Duncan, J. W. Enuiron. Sci. Technol. 1987,21, 466-475. (23) Stump, F.; Tejada, S.; Ray, W.; Dropkin, D.; Black, F.; Crews, W.; Snow, R.; Siudak, P.; Davis, C. 0.; Baker, L.; Perry, N. Atmos. Enuiron. 1989,23, 307-320. (24) Stump, F.; Tejada, S.; Ray, W.; Dropkin, D.; Black, F.; Snow, R.; Crews, W.; Siudak, P.; Davis, C. 0.;Carter, P. Atmos. Enuiron. 1990,24A, 2105-2112. (25) Bums, V. R.; Benson, J. D.; Hochhauser, A. M.; Koehl, W. J.; Kreucher, W. M.; Reuter, R. M. SAE Tech. Pap. Ser. 1991, No. 912320. (26) Hampton, C. V.; Pierson, W. R.; Harvey, T. M.; Updegrove, W. S.; Marano, R. S. Environ. Sci. Technol. 1982, 16, 287-298. (27) Hampton, C. V.; Pierson, W. R.; Schuetzle, D.; Harvey, T. M. Environ. Sei. Technol. 1983,17, 699-708. (28) Lonneman, W. A.; Seila, R. L.; Meeks, S. A. Enuiron. Sci. Technol. 1986,20, 790-796. (29) Dannecker, W.; Schroder, B.; Strechmann, H. Sci. Total Enuiron. 1990, 93, 293-300. (30) Ingalla, M. N.; Smith, L. R. Measurement of on-road vehicle emission factors in the California South Coast Air Basin. Volume I 1 unregulated emissions. Report to the Coordinating Research Council under Project SCAQS-1, Southwest Research Institute, San Antonio, TX, 1990. (31) Zweidinger, R. B.; Sigsby, J. E.; Tejada, S. B.; Stump, F. D.; Dropkin, D. L.; Ray, W. D.; Duncan, J. W. Enuiron. Sci. Technol. 1988, 22, 956-962. (32) Bailey, J. C.; Schmidl, B.; Williams, M. L. Atmos. Enuiron. 1990,24A, 43-52. (33) Aldehyde and reactive organic emissions from motor uehicles. Part 11-characterization of emissions from 1970 through 1973 model vehicles; APTD-1568b; U.S. Environmental Protection Agency: Ann Arbor, MI, 1973.
(34) Warner-Selph, M. A. Measurements of toxic exhaust emissions from gasoline-powdered light-duty vehicles. Report to the California Air Resources Board under Contract A6-198-32, Southwest Research Institute, San Antonio, TX, 1989. (35) Leppard, W. R.; Rapp, L. A.; Burns, V. R.; Gorse, R. A.; Knepper, J. C.; Koehl, W. J. SAE Tech. Pap. Ser. 1992, No. 920329. (36) Hirata, A. A. South Coast Recycled Auto Project (SCRAP). Unocal Corp., Los Angeles, CA, personal communication, 1990. (37) Volatile organic compound species data manual, 2nd ed.; EPA-450/4-80-015; NTIS PB81-119455; U.S. Environmental Protection Agency: Research Triangle Park, NC, 1980. (38) Bradow, R. L. Bradow & Associates, Raleigh, NC, personal communication, 1991. (39) Spicer, C. W. Composition and photochemical reactivity of turbine engine exhaust. Report ESL-TR-84-28 prepared for US. Air Force and Engineering Services Center, Batelle Columbus Laboratories, Columbus, OH, 1984. (40) Hoekman, S. K. Chevron Research and Technology Co., Richmond, CA, personal communication, 1991. (41) Rogozen, M. B.; Rapoport, R. D.; Shochet, A. Development and improvement of organic compound emissions inventories for California. Report to the California Air Resources Board under Contract AO-101-32, Science Applications International Corp., Hermosa Beach, CA, 1985. (42) Results of the 1988 architectural coatings sales survey. Stationary Source Division, California Air Resources Board, Sacramento, CA, 1991. (43) Ameritone products MSDS manual; Ameritone Paint Corp.: Long Beach, CA, 1988. (44) Material safety data sheets. Old Quaker Paint Co., Victorville, CA, 1991. (45) Material safety data sheets. Decratrend Paint Co., Industry, CA, 1988. (46) Material safety data sheets. Dunn-Edwards Co., Los Angeles, CA, 1986. (47) Materid safety data sheets. Sinclair Paint Co., Los Angeles, CA, 1990. (48) Gery, M. W.; Whitten, G. Z.; Killus, J. P.; Dodge, M. C. J. Geophys. Res. D 1989,94, 12925-12956. (49) Lawson, D. R. Research Division, California Air Resources Board, Sacramento, CA, personal communication, 1990. (50) Main, H. H.; Lurmann, F. W.; Roberta, P. T. Pollutant concentrations along the western boundary of the South Coast Air Basin. part I a review of existing data. Report to the South Coast Air Quality Management District. Sonoma Technology, Inc., Santa Rosa, CA, 1990. (51) Miller, M. S.; Friedlander, S. K.; Hidy, G. M. J. Colloid Interface Sci. 1972, 39, 165-176. (52) Friedlander, S. K. Enuiron. Sci. Technol. 1973, 7,235-240. (53) Mayrsohn, H.; Crabtree, J. H. Atmos. Enuiron. 1976,10, 137-143. (54) Nelson, P. F.; Quigley, S. M.; Smith, M. Y. Atmos. Enuiron. 1983,17,439-449. (55) Scheff, P. A.; Klevs, M. J. Enuiron. Eng. 1987, 113, 994-1005. (56) O’Shea, W. J.; Scheff, P. A. J. Air. Pollut. Control Assoc. 1988,38, 1020-1026. (57) Aronian, P. F.; Scheff,P. A.; Wadden, R. A. Atmos. Enuiron. 1989,23,911-920. (58) Case., G. R.; McRae, G. J. Enuiron. Sci. Technol. 1983,17, 129-139. (59) CMB7 user’s manual, Receptor model technical series; Volume 111. EPA-450/4-90-004; U.S. Environmental Protection Agency: Research Triangle Park, NC, 1990. (60) Chan, M.; Durkee, K. Southern California Air Quality Study B-Site operations. Report to the California Air Resources Board under Contract A5-196-32, Aerovironment Inc., Monrovia, CA, 1989. (61) Stockburger, L.; Knapp, K. T.; Ellestad, T. G. Overview and analysis of hydrocarbon samples during the summer Southern California Air Quality Study. Presented at the 82nd annual meeting of the Air and Waste Management Environ. Scl. Technol., Voi. 20,
No. 12, 1992 2407
Envlron. Sci. Technol, 1992, 26, 2408-2413
Association, Anaheim, CA, 1989; Paper 89-139.1. (62) Rasmussen, R. A. SCAQS hydrocarbon collection and analyses (Part I). Report to the California Air Resources Board under Contract A6-179-32, Biospherics Research Corp., Hillsboro, OR, 1990. (63) Finlayson-Pitts, B. J.; Pitts, J. N. Atmospheric
chemistry-fundamentals and experimental techniques, 1st ed.; John Wiley and Sons: New York, 1986;pp 313-316.
(64) Gorse, R. A. Environ. Sci. Technol. 1984, 18, 500-507. (65) Wesely, M. L. Atmos. Environ. 1989, 23, 1293-1304.
Received for review April 21,1992. Revised manuscript received July 28, 1992. Accepted August 3, 1992. This work was supported by the Electric Power Research Institute under Agreement RP3189-3.
Kinetics and Mechanism of the Reaction of H2S with Lepidocrocite Stefan Pelffer,*,+ Maria dos Santos Afonso,' Bernhard Wehrli,+ and Rene Gachtert
Lake Research Laboratory of EAWAGIETH, CH-6047 Kastanienbaum, Depto de Qdmica Inorgdnica, Anantica y Qdmica Flsica, Facultad de Ciencias Exactas y Naturales, Ciudad Universitarla Pabelldn 11, 1428 Buenos Aires, Argentina The initial reaction between hydrogen sulfide and the surface of lepidocrocite was studied in the pH range between 4 and 8.6 by monitoring the change of the emf of a pH2S sensor. The rate of H2S oxidation is pseudo first order with respect to H2S and shows a strong pH dependence with a maximum at pH 7. Two rate laws were derived: R 5FeS- and >FeHS. The overall rate law for the dissolution was derived as
R = k{>FeS-l + kl>FeHS)
(2)
{>FeS-)and {>FeHSjare surface concentrations (mol rn3.
0013-936X/92/0926-2408$03.00/0
0 1992 American Chemlcal Society
